Closed maxitg closed 5 years ago
By default, integration is done until final density is ~10^(-8) of original density:
In[] := AbsoluteTiming[ InflationEvolution[ 1/2 \[Phi]'[t]^2 - (1 - Cos[\[Phi][t]/10]), {{\[Phi], 30, 0}}, t]]
We can limit the bouncing to 10 to achieve similar results:
In[] := AbsoluteTiming[ InflationEvolution[ 1/2 \[Phi]'[t]^2 - (1 - Cos[\[Phi][t]/10]), {{\[Phi], 30, 0}}, t, "MaxBounceCount" -> 10, "ZeroDensityRelativeThreshold" -> 1.*^-5]]
We can achieve the same performance by setting "FinalDensityPrecisionGoal" to 5, but in that case there is a danger of it triggering during legitimate inflation slow-roll, which would never happen with bouncing:
In[] := AbsoluteTiming[ InflationEvolution[ 1/2 \[Phi]'[t]^2 - (1 - Cos[\[Phi][t]/10]), {{\[Phi], 30, 0}}, t, "FinalDensityPrecisionGoal" -> 5]]
Changes
Tests and commands
By default, integration is done until final density is ~10^(-8) of original density:
We can limit the bouncing to 10 to achieve similar results:
We can achieve the same performance by setting "FinalDensityPrecisionGoal" to 5, but in that case there is a danger of it triggering during legitimate inflation slow-roll, which would never happen with bouncing: